Hand detection image processing for a rescuing robot

Abstract

This report documents the summary of various approaches used for face-detection. Methods and algorithms developed for face-detection presently are studied to achieve the student’s final year project’s ultimate goal, which is to develop a Hand Detection method and algorithm for a rescuing robot, participating in RoboCup Singapore 2010. The method developed by Paul Viola and Michael Jones [1] – Boosted Cascade of Simple Haar Features is picked to be explored. Details on Viola & Jones’s contribution on their renowned object detection algorithm are documented in this report. A large training samples for hand-detection including images of human hands and mannequins’ hand were collected and used for classifier’s training via Viola & Jones. The details on training procedures are covered. Lastly, this report includes the detector’s performance and evaluation under various environment and lighting conditions. Future improvements on the hand-detector are also documented in this report.